Abstract: Thecode-phase divergences, which are absent for GPS, GLONASS, and Galileo, are commonlyfound in BDS geostationary (GEO), Inclined GeoSynchronous Orbit (IGSO) and Medium Earth Orbit (MEO) satellites. Several precise applications that use code observations are severely affected by these code biases; therefore, it is necessary to correct biases in BDS code observations. Since the BeiDou satellite-induced code bias is confirmed to be orbit type-, frequency-, and elevation-dependent, an improved code bias correction model for IGSO and MEO satellites based on a large amount of the data was developed. To obtain the best fitting results, we analyzed the effect of the number and distribution of stations and observation time on model estimation, and also considered the different influence of multipath at different elevations. A robust estimation method controlled the observation quality. A dataset from 18 stations during one year period in 2015 was employed to estimate the correction model for MEO satellites and four stations for IGSO satellite. To validate the improved correction model, the effect of the code bias on precise point positioning (PPP) before and after correction is analyzed and compared. Results show that systematic variations were eliminated more clearly after applying the improved correction model as compared to the traditional model. After correction, the positioning accuracy of PPP solution was improved and the convergence time decreasedshowing a better performance than results using thetraditional model as proposed by Wanninger and Beer.